Fair AI

Good examples matter for everyone.

01 · Read the story

Meet Mina and Sami

A school makes a voice helper, but it hears examples from only a few people. It sometimes misunderstands others. The students want to improve it.

02 · Discover the idea

What can we learn?

An AI model can be unfair or unreliable when its training examples or design leave out people or situations. Testing with varied examples helps us discover problems.

About this activity: This activity is a simple scenario quiz, not an audit of a real AI system.

03 · Compare mistakes

Does the helper work equally well?

Explore imaginary results for two speaking styles, each tested on ten examples. Consider what more testing might reveal.

Fairness experiment

Compare two tests

Which speaking style is understood less often?

Correct guesses out of 10

Compare the results.

Important: All numbers are invented. A small sample can reveal a concern, but cannot prove the cause or that a system is fair for everyone. Nobody is recorded.

04 · Review and explain

Choose, check, and explain.

There is no timer. If you make a mistake, learn from the explanation and try another question.

Question 1 of 30 correct

05 · Try it together

Family activity

Imagine making a voice helper for relatives of different ages. List kinds of voices, languages and noisy places to test, without recording anyone.

06 · Reflect

Talk about it

How can you make sure a helpful tool works for different people?

No answers are collected or sent anywhere. An adult can guide the conversation.